The search for biomarkers is a critical task in cancer research since understanding, diagnosis and prognosis of the illness can be greatly enhanced through them. This work represents an effort to make this search more effective and efficient when analyzing data from microarray experiments. The result is a novel method based on multiple criteria optimization that does not require the adjustment of parameters by the users, thereby allowing the consistent identification of important genes with potential biomarking characteristics. The high discrimination rate and accuracy of the method is evidenced through a series of applications in this work where potential biomarkers for cervix and colon cancer are presented.